Arduino Turns Off When Grounded To Shift Registers

I've got a problem that's perplexing me and I figured this would be the place to get some input. I've tried this with both a Duemilanove board and a Mega Uno 2560 board, and I get the same problem.

I have been slowly daisy chaining shift registers (74HC595s) which control LEDs for some research equipment I'm building. The registers get the clock, data, and latch signals from pins 8, 9, and 10 respectively.

The LEDs have a common ground which is attached to the Arduino, as do the shift registers. The shift registers also get their +5V source from the Arduino pin.

When I had 8 of these registers hooked up, it worked fine. When I added the 9th, every time I connect the common ground from the registers to the Arduino, the Arduino shuts down (the LEDs on the board cut off and it does nothing).

The ground for the LEDs doesn't cause any issue, and if I've tried the shift registers' ground in multiple locations, including just touching the wire for the LED ground, and I get the same result every time.

I've also tried this with both a battery and a wall-wart supply.

I'm pretty stumped, so if anyone has any advice, I'd really appreciate it.

So have you added any decoupling capacitors?
http://www.thebox.myzen.co.uk/Tutorial/De-coupling.html

I haven't.

Would that make a difference with the ground, though? The common +5V wire doesn't shut off the Arduino.

I've got some spare pins so I could break the shift registers down into sets of 5 (I'm going to end up with 15 in total) if that could solve the issue without necessitating waiting for new hardware to ship (where I live, the only place to go for hardware is Radioshack, and that place makes my blood boil).

I'm open to suggestion.

When I added the 9th, every time I connect the common ground from the registers to the Arduino, the Arduino shuts down

Are you using the same 9th chip every time? i.e. could this be a defective IC?

I don’t know why I didn’t think to switch out the chip. I even put them all in with sockets just in case one died on me. >_<

I’ll do that as soon as I get back to my lab tomorrow.

I haven't.

If you haven't got decoupling with 8 chips then you have problems. Even if there are bigger problems masking this one, it is simply not going to work without decoupling.

Good to know. I’ll try to get my hands on some capacitors ASAP.

Would that make a difference with the ground, though? The common +5V wire doesn't shut off the Arduino.

because the circuit isn't complete until ground is connected.

Are you powering everything off of the Arduino's regulator? You might be overloading the regulator or whatever power supply you are using.

The problem was the shift register itself. I've never seen an IC cause such catastrophic failure of a system as that.

because the circuit isn't complete until ground is connected.

Are you powering everything off of the Arduino's regulator? You might be overloading the regulator or whatever power supply you are using.

The only thing I can find from the data sheet regarding power consumption is: Low Power Dissipation: ICC = 4µA(MAX.) at TA=25°C.
I'm guessing that means it's consuming 4 micro amps, which, even with 15 shift registers, shouldn't cause an issue since the total load is a single LED plus the registers.

I read a bit more on decoupling capacitors and it makes a lot more sense now. I don't think I'm going to try to put them into this project, since it's a prototype, unless it becomes apparent that they're necessary. If everything was on a nice board, I'd do it in a heart beat, but I realized I'd be trying to shove in capacitors to a rat's nest of wires that are covering the registers and the entire thing is on some really shitty perf board.

I think I've got the problem hammered out, and I've been absolutely amazed by the patient, polite, and helpful responses from the community. What I'm working on is some visual psychophysical equipment, and if it works out well, I'll try to post some details in the finished projects section, as I'm hoping to keep it open source (assuming I don't run into any unexpected issues with my university).

Thanks again, folks.

I'm guessing that means it's consuming 4 micro amps, which, even with 15 shift registers, shouldn't cause an issue since the total load is a single LED plus the registers.

Depends on who's shift register you are using. For example, TI's is rated for 80uA. That's still a small amount of current. However, don't forget that whatever you are driving with those shift registers will increase your current draw.

I read a bit more on decoupling capacitors and it makes a lot more sense now. I don't think I'm going to try to put them into this project, since it's a prototype, unless it becomes apparent that they're necessary.

When you are spreading your power across that many ICs and components, decoupling capacitors are extremely important. Try and add caps anywhere you can to help your power distribution. Next time you build a prototype, keep in mind that generally a 0.1uF for each IC is generally a good idea. (And it is always easier to remove stuff later than to add it in.)

(And it is always easier to remove stuff later than to add it in.)

Yes but having built in some decoupling capacitors you would not want to take them all out, that is just madness.